scholarly journals Multi-Channel Target Speech Extraction with Channel Decorrelation and Target Speaker Adaptation

Author(s):  
Jiangyu Han ◽  
Xinyuan Zhou ◽  
Yanhua Long ◽  
Yijie Li
2007 ◽  
Vol 127 (6) ◽  
pp. 970-971
Author(s):  
Tadahiro Azetsu ◽  
Eiji Uchino ◽  
Noriaki Suetake

Author(s):  
Songxiang Liu ◽  
Jinghua Zhong ◽  
Lifa Sun ◽  
Xixin Wu ◽  
Xunying Liu ◽  
...  

2013 ◽  
Vol 37 (3) ◽  
pp. 611-620
Author(s):  
Ing-Jr Ding ◽  
Chih-Ta Yen

The Eigen-FLS approach using an eigenspace-based scheme for fast fuzzy logic system (FLS) establishments has been attempted successfully in speech pattern recognition. However, speech pattern recognition by Eigen-FLS will still encounter a dissatisfactory recognition performance when the collected data for eigen value calculations of the FLS eigenspace is scarce. To tackle this issue, this paper proposes two improved-versioned Eigen-FLS methods, incremental MLED Eigen-FLS and EigenMLLR-like Eigen-FLS, both of which use a linear interpolation scheme for properly adjusting the target speaker’s Eigen-FLS model derived from an FLS eigenspace. Developed incremental MLED Eigen-FLS and EigenMLLR-like Eigen-FLS are superior to conventional Eigen-FLS especially in the situation of insufficient data from the target speaker.


1992 ◽  
Vol 19 (2) ◽  
pp. 389-413 ◽  
Author(s):  
Dominique Bassano ◽  
Maya Hickmann ◽  
Christian Champaud

ABSTRACTThis study focuses on the development of epistemic modality, with particular attention to how French children evaluate the conditions of use for modal expressions marking certainty and uncertainty. Sixty children aged four, six and eight were shown films involving verbal interactions in which a target speaker accused another of having performed a deed. The analysis examine children's responses during a subsequent interview in which they were asked to attribute an epistemic attitude of certainty/uncertainty to the target speaker as a function of three factors: (a) whether he had witnessed the deed; (b) whether his accusation was modalized by the verbcroire(‘think/believe’); and (c) whether the accusation was true or false. The results show that the four-and six-year-olds attribute certainty more often than the eight-year-olds. This dissymmetry is accompanied by a developmental progression in children's conceptions of these modal categories, which change from a ‘realistic’ conception (mainly based on truth/falsity) at four years to an increasingly metalinguistic and relativized conception thereafter.


Author(s):  
Ryoichi Miyazaki ◽  
Hiroshi Saruwatari ◽  
Satoshi Nakamura ◽  
Kiyohiro Shikano ◽  
Kazunobu Kondo ◽  
...  
Keyword(s):  

2021 ◽  
pp. 101327
Author(s):  
Jiatong Shi ◽  
Chunlei Zhang ◽  
Chao Weng ◽  
Shinji Watanabe ◽  
Meng Yu ◽  
...  

2015 ◽  
Vol 32 (1) ◽  
pp. 19 ◽  
Author(s):  
Tetyana Sydorenko ◽  
Carson Maynard ◽  
Erin Guntly

The criteria by which raters judge pragmatic appropriateness of language learners’ speech acts are underexamined, especially when raters evaluate extended discourse. To shed more light on this process, the present study investigated what factors are salient to raters when scoring pragmatic appropriateness of extended request sequences, and which specific aspects of performance they attend to as appropriate or inappropriate. Three judges evaluated request sequences using a 6-point scale, marked appropriate and inappropriate elements of each request, and explained how they approached the rating of each response. It was found that all raters oriented to the appropriateness of a request sequence as a whole, paying attention not only to the request proper but also to all follow-up moves, including appreciation and closing. Additionally, raters oriented to the surrounding context: the same expressions, such as a specific appreciation statement, were rated as appropriate in some contexts and inappropriate in others. Raters also oriented to pragmatic competence broadly, paying attention not only to appropriate pragmatic strategies and expressions in a particular context, but also to such aspects as intonation and cultural knowledge. Finally, while native and near-native speaker tendencies were observed, target speaker norms were not. Implications for pragmatics teaching and assessment are discussed.Les critères selon lesquels les évaluateurs jugent la pertinence pragmatique des actes de langage d’apprenants de langue n’ont pas suffisamment fait l’objet d’études, notamment lors de l’évaluation de longues conversations. Pour éclairer davantage le processus, la présente étude a cherché à déterminer quels facteurs les évaluateurs jugent importants dans la pertinence pragmatique de séquences de requête étendues, et quels aspects spécifiques de la performance ils estiment appropriés ou pas. Trois juges ont évalué des séquences de requête selon une échelle de 6 points, ont indiqué les éléments appropriés et inappropriés de chaque requête et ont expliqué comment ils avaient abordé l’évaluation de chaque réponse. Les résultats indiquent que tous les évaluateurs jugeaient de la pertinence d’une séquence de requête dans son intégralité, portant attention non seulement à la requête comme telle mais aussi à toutes les démarches qui la suivaient, y compris le remerciement et la clôture. De plus, les évaluateurs tenaient compte du contexte : ils jugeaient qu’une même expression, une déclaration spécifique d’appréciation par exemple, était appropriée dans un contexte donné alors qu’elle ne l’était pas dans un autre. Ils ont également considéré la compétence pragmatique globale, notant, au delà des stratégies et des expressions pragmatiques appropriées dans un contexte donné, des aspects comme l’intonation et les connaissances culturelles. Finalement, si les évaluateurs ont observé des tendances de locuteurs natifs ou quasi-natifs, on ne peut en dire autant des normes de la langue cible. On discute des incidences de l’étude sur l’enseignement et l’évaluation des compétences pragmatiques.


eLife ◽  
2017 ◽  
Vol 6 ◽  
Author(s):  
Antonio Elia Forte ◽  
Octave Etard ◽  
Tobias Reichenbach

Humans excel at selectively listening to a target speaker in background noise such as competing voices. While the encoding of speech in the auditory cortex is modulated by selective attention, it remains debated whether such modulation occurs already in subcortical auditory structures. Investigating the contribution of the human brainstem to attention has, in particular, been hindered by the tiny amplitude of the brainstem response. Its measurement normally requires a large number of repetitions of the same short sound stimuli, which may lead to a loss of attention and to neural adaptation. Here we develop a mathematical method to measure the auditory brainstem response to running speech, an acoustic stimulus that does not repeat and that has a high ecological validity. We employ this method to assess the brainstem's activity when a subject listens to one of two competing speakers, and show that the brainstem response is consistently modulated by attention.


Sign in / Sign up

Export Citation Format

Share Document